International Journal of Food Microbiology 131 (2009) 120127

Contents lists available at ScienceDirect

International Journal of Food Microbiology

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j f o o d m i c r o

Effects of temperature, pH and sugar concentration on the growth parameters of

Saccharomyces cerevisiae, S. kudriavzevii and their interspecic hybrid
F. No Arroyo-Lpez a, Sandi Orli a,c, Amparo Querol b, Eladio Barrio a,
a
b
c

Institut Cavanilles de Biodiversitat i Biologia Evolutiva, Universitat de Valncia, Edici d'Instituts, Parc Cientc de Paterna, P.O. Box 22085, E-46071 Valncia, Spain
Departamento de Biotecnologa de Alimentos, Instituto de Agroqumica y Tecnologa de los Alimentos, CSIC, P.O. Box 73, E-46100 Burjassot, Valencia, Spain
Department of Microbiology, Faculty of Agriculture, University of Zagreb, Svetosinuska 25, 10 000 Zagreb, Croatia

a r t i c l e

i n f o

Article history:
Received 23 July 2008
Received in revised form 8 January 2009
Accepted 25 January 2009
Keywords:
Central composite design
Hybrid yeast
Response surface
Saccharomyces cerevisiae
Saccharomyces kudriavzevii
Wine fermentation

a b s t r a c t
The effects of temperature, pH and sugar concentration (50% glucose + 50% fructose) on the growth
parameters of Saccharomyces cerevisiae T73, S. kudriavzevii IFO 1802T and the hybrid strain S. cerevisiae
S. kudriavzevii W27 were studied by means of response surface methodology based in a central composite
circumscribed design. Lag phase could not be properly modelled in the wine model system, where yeasts
started the fermentation in few hours after inoculation. In the case of the maximum specic growth rate
(max), the temperature was the most important variable for three yeasts, although the effects of sugar
concentration (in T73 and W27) and pH (W27 and 1802) were also signicant (p b 0.05). The only retained
interaction was between the variables temperature and pH for yeast 1802. The polynomial equations built for
max were used both to assess the behaviour of yeasts as a function of the factors and to predict their growth.
In the case of temperature, the proles obtained by the equations showed that response of the hybrid W27
was similar to T73 and different to 1802. When pH was the factor under study, the response of the hybrid
W27 was closer to 1802 than yeast T73. For sugar concentration, the response of the hybrid W27 was similar
to T73 but different to 1802. To the best of our knowledge, this is the rst time that predictive models are
used to assess and compare the response of a hybrid strain with respect to its parental species. The
information obtained could also be useful to estimate the possible effect of climatic change on yeast growth.
2009 Elsevier B.V. All rights reserved.

1. Introduction
Yeasts play a prominent role in wine fermentations, which can
strongly affect the quality and avour of the nal product (Querol
and Fleet, 2006). Among several yeasts, Saccharomyces cerevisiae and
S. bayanus var. uvarum are the most important species present during
the fermentation process (Pretorius, 2000; Querol and Fleet, 2006).
Recently, interspecic hybrid strains between Saccharomyces species
have been described as involved in wine fermentations. Gonzlez et al.
(2006) described wine yeast hybrids between the species S. cerevisiae
Saccharomyces kudriavzevii and S. cerevisiae S. bayanus. Several strains
selected as commercial wine yeasts also resulted to be Saccharomyces
hybrids (Gonzlez et al., 2006; Bradbury et al., 2006), for instance the
hybrid S. cerevisiae S. kudriavzevii Lalvin W27. Therefore, hybrid strains
appear as well adapted to the stress conditions (low pH, high sugar
concentration and ethanol content) occurring during wine fermentations (Belloch et al., 2008), and their enological characterization
conrmed their interesting properties according to the new trends in
winemaking (Gonzlez et al., 2007).

Corresponding author. Tel.: +34 963543667; fax: +34 963543670.

E-mail address: eladio.barrio@uv.es (E. Barrio).
0168-1605/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.ijfoodmicro.2009.01.035

Different factors can affect the course of the fermentation, inuencing the ecology and adaptation of the microbiota present. The
temperature is a variable that directly affects the growth rate of the
microorganisms (Charoenchai et al., 1998), and the nal composition
of wine (Torija et al., 2003). Another signicant variable is the concentration of fermentable sugars in grape musts, ranging between 125
and 250 g/L (Fleet and Heard, 1993). It is likely that the initial
concentrations of glucose and fructose (main grape sugars) will
selectively inuence the species and strains of yeast present during
fermentation. Must pH, ranging from 2.75 to 4.25, is also considered
an important factor for the survival and growth of yeasts (Fleet
and Heard, 1993). Due to climatic change, glucose and fructose are
increasing their concentrations in grapes meanwhile the acidity
decreases, affecting the global wine quality (Jones et al., 2005). This
fact originates musts with a higher initial amount of fermentable
sugars and higher pH. Hence, these factors must be studied with more
detail, especially the interactions between them and their inuence
on microorganisms.
Several studies have modelled the wine fermentation process
(Malherbe et al., 2004; Colombi et al., 2005; Coleman et al., 2007),
whereas other works focused on studying the inuence of environmental variables on the microorganisms involved in wine fermentations. In this context, Charoenchai et al. (1998) reported the effects of

F.N. Arroyo-Lpez et al. / International Journal of Food Microbiology 131 (2009) 120127

temperature, pH, and sugar concentration on the growth rates and

cell biomass of several wine yeasts such as Kloeckera apiculata, Torulaspora delbrueckii, Pichia anomala, S. cerevisiae and various Candida
species. Medawar et al. (2003) modelled the lag phase of Brettanomyces intermedius i100 as a function of ethanol content in laboratory
medium, whereas D'Amato et al. (2006) showed the inuence of
temperature, ammonium and glucose concentrations on S. cerevisiae
growth in a synthetic must. Finally, Serra et al. (2005) studied the
effects of temperature and pH on the growth of S. bayanus var. uvarum
P3, S. cerevisiae VL3C and their interspecic hybrid in order to use
hybrids as active dry yeasts. However, more information is still necessary to understand the behaviour of yeasts during wine fermentation.
The prediction of the kinetic of the fermentation process in relation to
the properties of the grape must is of great interest for wine industry
(D'Amato et al., 2006).
Response surface (RS) is a very useful tool which has been
previously applied to estimate the effects of environmental variables
on yeast growth (Sorensen and Jakobsen, 1997; D'Amato et al., 2006;
Arroyo-Lpez et al., 2006). This methodology has widely been used in
predictive microbiology as a secondary model (McMeekin et al., 1993)
to predict the microorganism response to environmental changes.
Particularly, central composite designs with star points are very useful
because they provide rotatability, high quality predictions over the
entire design space and low number of experimental runs (Myers and
Montgomery, 2002).
The aims of this paper are: a) to estimate the effects of temperature, pH and sugar concentration (50% glucose + 50% fructose) on the
yeasts S. cerevisiae T73, S. kudriavzevii IFO 1802T, and the hybrid strain
S. cerevisiae S. kudriavzevii W27, obtaining at the same time valuable
information about the possible effect of climatic change on their
growth, b) to use mathematical modelling obtained from RS
methodology to predict the response of these microorganisms under
different combinations of the environmental variables, and c) to
assess and compare the response of the hybrid strain with respect to
representative strains of its parental species.
2. Materials and methods
2.1. Yeast strains
The yeasts used in this study were S. cerevisiae Lalvin T73 (abbreviated as T73), S. kudriavzevii IFO 1802T (hereafter 1802) and the hybrid
strain S. cerevisiae S. kudriavzevii Lalvin W27 (henceforth W27). T73
was selected years ago in our laboratory from a wine fermentation in
Alicante, Spain (Querol et al., 1992), and is commercialised as Lalvin T73
(Lallemand Inc., Montreal, Canada). The hybrid W27 was also isolated
from wine fermentations in Wdenswill, Switzerland (Schtz and
Gafner, 1994), and is also commercialised (Lallemand Inc.). Finally,
S. kudriavzevii has never been described in wine fermentation. For this
reason we have used the type strain of this species, IFO 1802T, isolated
from decayed leaves in Japan (Naumov et al., 2000). The strains T73 and
1802 were used as representatives of the parental species involved in the
formation of the hybrid W27, because as a natural hybrid their real
parental strains are unknown.

121

2.3. Growth medium preparation

Fermentation experiments were carried out in a complex synthetic
medium MS300 miming a standard natural must previously described
by Bely et al. (1990). Natural musts show a variable composition
among different years that can inuence the yeast growth. For this
reason, a dened synthetic must was chosen in this work as the most
appropriate growth medium to overcome this variation. Firstly the
sugar concentration (50% glucose + 50% fructose) was adjusted in
distilled water according to the experimental design and heated at
100 C for 15 min to prevent sugar caramelization. Stocks for the other
components of the medium (mineral salts, vitamins, amino acids and
anaerobic factors) were previously sterilized by ltration (0.2 m) and
then added to the basal medium in the appropriate concentration
(Rossignol et al., 2003). Finally the pH was adjusted by aseptically
adding tartaric acid (85%, wt/vol) according to the experimental
design. We chose this organic acid because it is a compound normally
found in grapes and wines and it is very rarely metabolized by
ascomycetous yeasts. Sterile glass bottles (60 mL of volume) were
lled with 55 mL of synthetic must and independently inoculated
with 50 L of the corresponding yeast saline suspension to reach an
initial concentration of inoculum of about 4.50 0.21 log10CFU/mL
for T73, 4.36 0.32 log10CFU/mL for the hybrid W27, and 4.15 0.29
log10CFU/mL for 1802. Bottles were incubated without shaken.
2.4. Experimental design
The effects of sugar concentration (50% glucose + 50% fructose),
temperature, and pH on yeasts growth were tested using a central
composite circumscribed design with three repetitions in the centre to
account for pure error. The design was generated with Statistica 6.0
software package (StatSoft, Tulsa, OK, USA) and environmental
variables levels were established into a range of conditions usually
found in wine fermentations. Star points were situated at 1.68 from
the centre to account for rotatability, which permits to extract the
same amount of information (and consequently make predictions
with the same precision) in all directions of the tted surface (Myers
and Montgomery, 2002). The full central composite circumscribed
design, based on three basic principles of an ideal experimental
design, primarily consists of (1) a complete 2n factorial design, where
n is the number of test variables, (2) n0 number of centre points and
(3) start points on the axis of each design variable at a distance of 1.68
from the design centre. Hence, the total number of design points was
N = 2n + 2n +n0, in our case 17 treatments. For statistical calculations
the variables were coded according to the equation:
Coded variable = physical value 1 = 2high value + low value
= 1 = 2high value low value
1
Relationships between coded and physical levels are shown in
Table 1. All experiments were randomly performed in duplicate. The
full experimental design was independently run for each yeast and
monitored for at least two weeks.

2.2. Inoculum preparation

Single colonies from pure cultures of each species were inoculated
separately into 5 mL of Yeast-Malt-peptone-glucose broth medium
(YM, Difco, Becton and Dickinson Company, Sparks, USA) and
then incubated at 25 C for 24 h. The initial pH and glucose concentration of the medium was 5.5 0.2 and 10 g/L, respectively. After
this period, tubes were centrifuged, the pellets washed with sterile
saline solution (0.9% NaCl, wt/vol) centrifuged and re-suspended
again in sterile saline solution to obtain a concentration of about 7
log10 CFU/mL.

Table 1
Relationship among physical and coded values of environmental variables used in the
central composite circumscribed design.
Point

Temperature (C)
pH
Sugar concentrationa (g/L)
a

Coded values
Star

Low

Centre

High

Star

1.68

1.00

0.00

+ 1.00

+ 1.68

13.9
2.24
116

18.0
2.75
150

24.0
3.50
200

30.0
4.25
250

34.1
4.76
284

(50% glucose + 50% fructose).

122

F.N. Arroyo-Lpez et al. / International Journal of Food Microbiology 131 (2009) 120127

2.5. Enumeration of cells

Samples were taken from the fermentations at variable time
intervals, diluted in sterile saline solution and plated onto YM agar
plates. Sampling times were established according to the yeast growth
conditions so that the total number of samples ranged from 8 to 15
and were distributed through the different phases of the yeast growth
(lag, exponential and stationary). All plates were incubated aerobically at 25 C for 48 h. Counts were expressed as CFU/mL.

(temperature (T), pH and sugar concentration (S), respectively). For

the main effects, regression coefcients can be interpreted as the
increase or decrease (depending of the positive or negative coefcient
sign) in the response when the factor changes one unit. Analysis of the
RS was made using the Experimental Design module of the Statistica
6.0 software package, using the pure error, derived from repetitions in
the centre, as option in the corresponding ANOVAs. Model performance was checked by the lack of t test and the determination
coefcient R2 (percentage of variability in the response that can be
explained by the model).

2.6. Primary model

2.8. Model validation
Growth parameters (maximum specic growth rate and lag phase
duration) were calculated from each treatment by directly tting plate
count versus time to the reparametized Gompertz equation proposed
by Zwietering et al. (1990):


y = D exp exp max e = D t + 1

where y = ln(Nt/N0), N0 is the initial population (CFU/mL) and Nt is the

population at time t; D = ln(N/N0) is the maximum population value
reached with N as the asymptotic maximum, max is the maximum
specic growth rate (h 1), and the lag phase period (h). Growth data
from each treatment and yeast were tted by a non-linear regression
procedure, minimizing the sum of squares of the difference between
experimental data and the tted model, i.e., loss function (observed
predicted)2. This task was accomplished using the non-linear module of
the Statistica 6.0 software package and its Quasi-Newton option.
2.7. Secondary model
Lag phase () and maximum specic growth rate (max) were
subsequently adjusted to a response surface of the general form:
2

For the model validation, a new series of experiments were carried

out in synthetic must. Each treatment was a combination of all studied
variables. Variable levels were chosen randomly within the range of
the values used in the design (interpolation region). Validation treatments were: 1) 21.0 C, pH 3.00, 225 g/L sugar (coded values, 0.50,
0.66, and 0.50, for temperature, pH and sugar concentration,
respectively); 2) 21.0 C, pH 4.00, 175 g/L sugar (coded values,
0.50, 0.66, 0.50); 3) 27.0 C, pH 3.00, 225 g/L sugar (coded values,
0.50, 0.66, 0.50); and 4) 27.0 C, pH 4.00, 175 g/L sugar (coded
values 0.50, 0.66, 0.50). Experimental max values were compared
to those predicted by the RS equations. To give a quantitative measure
of the models performance, the accuracy (A) and bias factors (B) were
calculated as described by Baranyi et al. (1999). The accuracy factor is
based on mean square differences, while the bias factor is based on the
arithmetical mean of the differences. The indexes %D (percent
discrepancy between the predictions and observations), and %B
(percent bias) were also obtained (Baranyi et al., 1999).
3. Results
3.1. Effects of environmental variables on yeast growth parameters

Y = Qo + Q1 X1 + Q11 X1 + Q2 X2 + Q22 X2 + Q3 X3 + Q33 X3 + Q12 X1 X2 3

+ Q13 X1 X3 + Q23 X2 X3 + e:

where Y is the growth parameter modelled, o is the mean/intercept

term, i are the coefcients to be estimated during the RS tting (1 is
the coefcient for the linear effect of X1, 11 for the quadratic effect of
X1, 12 for the interaction between variables X1 and X2, and so on) and
is the term for error. X1, X2, and X3 are the environmental variables

Yeasts had a very short lag period under the conditions included in
the experimental design (Table 2). This parameter ranged from 0.89 to
8.39 h in the case of T37, from 0.92 to 10.1 h for the hybrid W27, and
from 0.30 to 16.7 h in case of yeast 1802 (excluding run 10 in which no
growth was observed after 170 h). Modelling lag phase in the wine
model system showed difculties. The percentages of variability in the
response (R2) that could be explained by the models for were low,

Table 2
Treatments included in the central composite circumscribed design and growth parameters (lag phase () and maximum specic growth rate (max)) obtained for the yeasts
Saccharomyces cerevisiae T73, S. kudriavzevii IFO 1802T and the hybrid strain S. cerevisiae S. kudriavzevii W27.
max (h 1)

(h)

Coded values
Run

pH

T73

W27

1802

T73

W27

1802

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.68
1.68
0.00
0.00
0.00
0.00
0.00
0.00
0.00

1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
0.00
0.00
1.68
1.68
0.00
0.00
0.00
0.00
0.00

1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
0.00
0.00
0.00
0.00
1.68
1.68
0.00
0.00
0.00

4.50 (0.86)
3.12 (1.20)
4.93 (2.48)
1.02 (0.44)
2.61 (1.41)
2.23 (1.09)
2.32 (0.88)
4.13 (1.23)
5.38 (1.87)
0.89 (1.18)
8.39 (1.95)
3.09 (1.22)
3.84 (1.57)
2.02 (2.51)
1.87 (0.82)
2.57 (1.39)
1.09 (0.36)

3.14 (1.89)
0.92 (0.15)
8.45 (1.49)
3.67 (1.60)
3.08 (0.58)
2.47 (0.68)
4.34 (0.81)
3.53 (0.91)
5.28 (1.96)
3.46 (0.78)
10.1 (1.05)
3.45 (0.67)
3.27 (1.49)
2.39 (1.24)
4.32 (0.50)
4.36 (0.85)
4.07 (0.80)

0.89 (0.94)
0.93 (0.82)
4.04 (2.96)
1.92 (1.50)
3.89 (1.17)
16.7 (6.08)
4.64 (1.03)
5.71 (0.69)
3.95 (1.90)
170 (0.00)a
0.30 (1.05)
6.31 (0.95)
2.79 (0.73)
4.91 (1.39)
4.49 (0.41)
4.42 (0.77)
3.94 (0.54)

0.113 (0.026)
0.096 (0.017)
0.129 (0.019)
0.094 (0.012)
0.339 (0.095)
0.353 (0.097)
0.422 (0.054)
0.441 (0.088)
0.082 (0.007)
0.293 (0.040)
0.218 (0.034)
0.317 (0.055)
0.480 (0.102)
0.162 (0.022)
0.224 (0.037)
0.241 (0.038)
0.198 (0.024)

0.157 (0.026)
0.150 (0.010)
0.177 (0.014)
0.151 (0.018)
0.459 (0.043)
0.415 (0.042)
0.644 (0.117)
0.465 (0.077)
0.096 (0.006)
0.557 (0.089)
0.046 (0.008)
0.334 (0.032)
0.421 (0.098)
0.272 (0.030)
0.324 (0.023)
0.361 (0.051)
0.319 (0.031)

0.142 (0.010)
0.137 (0.015)
0.161 (0.025)
0.161 (0.012)
0.143 (0.023)
0.119 (0.044)
0.348 (0.054)
0.275 (0.022)
0.117 (0.009)
0.000 (0.000)a
0.143 (0.037)
0.522 (0.099)
0.388 (0.045)
0.394 (0.087)
0.354 (0.024)
0.383 (0.054)
0.337 (0.028)

Note: Standard deviation calculated from duplicate experiments in parenthesis.

T, temperature; S (sugar concentration).
a
No growth was observed in this treatment for yeast 1802 and max and were set up at 0.000 h 1 and 170 h for analysis, respectively.

F.N. Arroyo-Lpez et al. / International Journal of Food Microbiology 131 (2009) 120127
Table 3
Regression coefcients estimated by means of the ANOVA analysis for the response
factor Y (maximum specic growth rate, h 1) in function of temperature (T), pH and
sugar concentration (S) for the yeasts Saccharomyces cerevisiae T73, S. kudriavzevii IFO
1802T and the hybrid S. cerevisiae S. kudriavzevii W27.
Regression coefcient

Value

Standard deviation

p-value

Model for yeast T73

0 (Mean/Inter)
1 (linear T)
11 (quadratic T)
2 (linear pH)
22 (quadratic pH)
3 (linear S)
33 (quadratic S)
12 (interaction T pH)
13 (interaction T S)
23 (interaction pH S)

R2 = 0.812
0.220
0.108
0.014
0.025
0.013
0.040
0.032
0.019
0.010
0.001

0.013
0.006
0.006
0.006
0.007
0.006
0.007
0.008
0.008
0.008

Model for yeast W27

0 (Mean/Inter)
1 (linear T)
11 (quadratic T)
2 (linear pH)
22 (quadratic pH)
3 (linear S)
33 (quadratic S)
12 (interaction T pH)
13 (interaction T S)
23 (interaction pH S)

R2 = 0.952
0.331
0.155
0.006
0.054
0.041
0.037
0.013
0.026
0.023
0.019

0.013
0.006
0.006
0.006
0.006
0.006
0.007
0.008
0.008
0.008

Model for yeast 1802

0 (Mean/Inter)
1 (linear T)
11 (quadratic T)
2 (linear pH)
22 (quadratic pH)
3 (linear S)
33 (quadratic S)
12 (interaction T pH)
13 (interaction T S)
23 (interaction pH S)

R2 = 0.854
0.362
0.006
0.121
0.076
0.024
0.006
0.003
0.039
0.011
0.005

0.013
0.006
0.006
0.006
0.007
0.006
0.006
0.008
0.008
0.008

0.003
0.003
0.175
0.054
0.180
a
0.023
a
0.041
0.139
0.325
0.861
a

123

their respective coefcients. In the case of temperature, max increased

its value when this environmental variable ranged from 13.9 to
34.1 C. One unit increase in temperature (in other words, the factor is
moved from 0 to + 1, change of 6 C) may, theoretically, increase max
in 0.108 h 1. The value of the coefcient for the sugar content can be
interpreted in a similar way: an increase of +1 in sugar concentration
(or 50 g/L in physical level) originates a change of 0.040 (h 1) in
the linear term and +0.032 (h 1) in the square term on the response,
showing a net decrease. No signicant coefcients were found for pH
(Table 3). However, these effects can be better studied by means of the
proles obtained by its respective polynomial equation, plotted in
Fig. 1. These graphs depict the changes in the parameter modelled as

0.001
0.001
0.439
a
0.012
a
0.025
a
0.026
0.184
0.080
0.099
0.140
a

0.001
0.416
0.003
a
0.006
0.072
0.396
0.650
a
0.040
0.296
0.572
a

Signicant coefcients (p-value 0.050).

with 67.9% for T73, 62.7% for yeast 1802, and 45.3% for the hybrid W27.
In addition, the lack of t was signicant (p b 0.05) for the ANOVA of
the hybrid W27 and yeast 1802, and only the model for T73 had a non
signicant lack of t (p N 0.143). Diverse transformations for (1/,
log10, and 1/) were also tted by RSM, but with similar results. In
any case, the experimental values recorded for showed that these
yeasts can start the fermentation in few hours after inoculation even
at low values of temperature and pH and high sugar concentration
(Table 2). Only run 10 for yeast 1802 (34.1 C, pH 3.50, 200 g/L)
showed no growth at the end of the experimental period, indicating
that such conditions were inhibitory for this microorganism.
Table 2 also shows the max obtained for the different yeasts
according to the experimental design. This parameter ranged from
0.082 to 0.480 h 1 in the case of T73, from 0.046 to 0.644 h 1 for the
hybrid W27, and from 0.000 to 0.522 h 1 for yeast 1802. The regression coefcients estimated for the three yeasts and their signicances,
deduced from the ANOVA analysis of regression, are shown in Table 3.
All models had a non-signicant lack of t (p N 0.05). In addition, the
variance in the response that was explained by these models was
always high: 81.2% for T73, 85.4% for yeast 1802 and 95.2% for the
hybrid W27. This way, the models can be considered appropriated to
describe the effects of environmental variables on max.
A regression coefcient that was signicant (p-value 0.05) for
the three models was 0, which represents the mean/intercept term.
In other words, it is the value of max at the reference point 24.0 C,
pH 3.50, 200 g/L (0,0,0 in coded values). Other signicant coefcients
for yeast T73 were also the linear effect of temperature, and the linear
and quadratic effects of sugar concentration (Table 3). The effect of the
environmental variables on max of T73 can be studied by means of

Fig. 1. Changes in max (h 1) obtained by the polynomial equation of yeast Saccharomyces cerevisiae T73 as a function of a) temperature, b) pH, and c) sugar concentration,
with all other factors held constant at the central point (24.0 C for temperature, 3.50 for
pH and 200 g/L for sugar).

124

F.N. Arroyo-Lpez et al. / International Journal of Food Microbiology 131 (2009) 120127

the factor moves along its axis, with all other factors held constant at
the central point (0,0). Within the interval assayed, max always
increased with temperature (Fig. 1a). On the contrary, max decreased
progressively as the sugar concentration increases and reaches a
minimum around 220 g/L (Fig. 1c). From Fig. 1b, it is also clear that the
initial pH did not affect growth of this species, although a slight
increase of max was noticed when pH rose.
In the case of hybrid W27, the signicant coefcients were: the
linear effect of temperature, pH and sugar concentration, as well as the
quadratic term of pH (Table 3). This means that max increased
linearly when temperature ranged from 13.9 to 34.1 C due to its
positive term (0.155). The negative value of the linear effect of sugar is
indicative of an inverse relationship: an increase of 1 unit in sugar
decreases max by 0.037 h 1. Interpretation of the effect of pH is more
complex due to the combined effect of linear and quadratic terms. In
fact, the response may have an increase because of its positive linear
term (0.054) and a decrease due to the negative value of its quadratic
effect ( 0.041). As previously, these effects are better showed by the
respective proles obtained for the factors (Fig. 2). The effects of
temperature and sugars are fairly clear (Fig. 2a and c) and the overall
effect of pH is now more evident. max increased progressively from
pH 2.24 up to 3.80, where reaches a maximum (Fig. 2b).
The signicant coefcients for strain 1802 were: the linear effect of
pH, the quadratic effect of temperature, and the interaction temperature pH (Table 3). Interpretation of coefcients is similar to the
previous yeasts. The presence of signicant quadratic coefcient for
temperature is indicative of a curvature on this parameter. The
positive value of the interactive term between temperature and
pH (0.039) may indicate a potential synergistic effect between both
factors although no physiological interpretation can be deduced, given
de empirical character of the model. For estimation purposes, an
increase of 1 unit in the interaction temperature pH increases max in
0.039 h 1. The obtained proles clarify the effects. The effect of temperature showed a convex shape and an optimum in max was reached
around 24 C (Fig. 3a). According to Fig. 3b, max increased as pH
moved from 2.24 up to 4.76 (positive sign for its coefcient). Finally,
max did not change as the sugar concentration increased (Fig. 3c).
Overall, the temperature was, apparently, the variable with the
greater effect on yeast growth as may be deduced from the comparative study of Figs. 13, in which the changes in temperature always
produced the higher variations on max. Within the experimental
region, the effect of temperature was never related to the level of
sugar but showed a clear relationship with pH (signicant interaction
pH temperature) only in the case of yeast 1802, which may indicate
that it was strain dependent.
The RS equations obtained for the three yeasts as a function of
temperature, pH and sugar concentration can be deduced from
Table 3, by replacing the appropriate terms in Eq. (3). To reduce the
complexity of the equations, only signicant coefcients can be used.
These polynomial equations, in coded terms, can be used by the
industry to estimate the response (parameter modelled) of the three
strains as a function of diverse combinations of environmental variables within the experimental range (interpolation region). Thus, the
model may be useful for winemakers. For this purpose, it could be
advisable to transform physical values into coded terms according
to Eq. (1), and work with the coded polynomial equations to make
predictions.
3.2. Validation results
Predictions of the models for the three yeasts were good. The best
indexes for max were obtained for yeasts T73 (A = 1.04, B = 0.98) and
the hybrid W27 (A = 1.06, B = 0.98) with a pert cent discrepancies
between predictions and observations of 4.80% and 6.30%, respectively. The bias factor (B) for both yeasts showed values very close to 1,
with a slight tendency in the model to predict slower growth than the

Fig. 2. Changes in max (h 1) obtained by the polynomial equation of the hybrid

S. cerevisiae S. kudriavzevii W27 as a function of a) temperature, b) pH, and c) sugar
concentration, with all other factor held constant at the central point (24.0 C for
temperature, 3.50 for pH and 200 g/L for sugar).

observations (%B = 0.78 for T73 and 0.63 for hybrid W27). In the
case of yeast 1802, the indexes were slightly worse (A = 1.17, B = 1.14),
with a percent discrepancy of 17.5% and a bias factor N1, indicative that
the models predict faster growth than the observations (%B = + 5.82).
3.3. Assessing the response of the hybrid strain respect to its parental
species
The range checked for the three environmental variables (temperature 13.934.1 C, pH 2.244.76, and sugar concentration 116
284 g/L) is usually found in wine fermentations. Thus, from the results

F.N. Arroyo-Lpez et al. / International Journal of Food Microbiology 131 (2009) 120127

125

obtained in this study, we can assess and compare the response of the
hybrid strain W27 with respect to its parental species under a
simulated wine environment.
As mentioned above, values were very low, showing that all
yeasts had a short adaptation time to wine environmental conditions
even when the pre-incubation of the inoculums were carried out
in a medium with a higher pH (5.5) and lower sugar concentration
(10 g/L).
However, the response was different when the parameter analysed
was max. A comparison for this parameter with respect to temperature for yeasts T73, W27 and 1802 can be carried out by means of the
proles obtained for max using their respective RS equations, xing

Fig. 4. Response surface for max (h 1) of S. kudriavzevii IFO 1802T as a function of

temperature and pH for a sugar concentration of 116 g/L (50% glucose + 50% fructose).

the pH and sugar concentration at 3.50 and 200 g/L respectively (see
Figs. 1a3a). Strain T73 and the hybrid W27 had a linear response in
the range studied (13.934.1 C) with the maximum value for both
yeasts at 34.1 C, while 1802 showed a quadratic response with a
maximum around 24 C. In the case of pH, the proles obtained for
max, xing the temperature and sugar concentration at 24.0 C and
200 g/L, showed that the response for yeasts 1802 and the hybrid
W27 was more similar between them than with respect to T73
(Figs. 1b3b). All yeasts increased max when pH increased, but the
hybrid W27 showed an evident quadratic effect for this variable with
an optimum around 3.54.0 (see Table 3 and Fig. 2b). Yeasts 1802 and
T73 also showed a slight curvature for the pH, but their quadratic
effects were not signicant according to the corresponding ANOVAs
(Table 3). The proles obtained for max as a function of sugar
concentration xing the temperature and pH at 24.0 C and 3.50
(Figs. 1c3c), showed that max of 1802 was practically not affected by
the sugar concentration in the range studied (116284 g/L). However,
for yeasts W27 and T73 the response was similar, and max decreased
when sugar concentration increased, with a marked quadratic effect in
the case of T73 (see Table 3).
Results above commented were obtained xing two variables at
the centre of the experimental design (0, 0) and changing the other
variable along its range (1.68 to +1.68). They must be interpreted
as the different response of the yeasts only under these levels. But it is
also very interesting to nd the optimal growth conditions inside the
experimental region for the combinations of the three environmental
variables. These values can be obtained by optimizing mathematically
the proles for predicted values of max using their respective equations. They were attained at: a) 34.1 C, pH 4.76 and 284 g/L (50%
glucose + 50% fructose) in the case of yeast T73, with a predicted max
of 0.551 0.132 h 1; b) 34.1 C, pH 4.76 and 200 g/L for the hybrid
W27 with a predicted max of 0.660 0.151 h 1; and c) 24.0 C, pH 4.76
and 116 g/L for the yeast 1802 with a predicted max of 0.438 0.152 h 1.
Fig. 4 shows as an example the RS for max of yeast 1802 as a function
of temperature and pH for a sugar concentration of 116 g/L. It shows a
clear optimum around 24.0 C and pH 4.76 in agreement with the data
deduced from the previous proles for this yeast. The quadratic effect
along the temperature axis is also clear.
Fig. 3. Changes in max (h 1) obtained by the polynomial equation of yeast
S. kudriavzevii IFO 1802T as a function of a) temperature, b) pH, and c) sugar concentration, with all other factor held constant at the central point (24.0 C for temperature,
3.50 for pH and 200 g/L for sugar).

4. Discussion
In this work, RSM based in a central composite design proved to be
a very useful tool to assess the response of different yeasts under

126

F.N. Arroyo-Lpez et al. / International Journal of Food Microbiology 131 (2009) 120127

varying conditions of temperature, pH and sugar concentrations

usually found during wine fermentations. The methodology was
performed following a complete procedure of predictive microbiology
(experimental design, primary model, secondary model and validation) and was applied to wine yeast strains. Validation indexes
obtained from independent experiments showed that RS equations
can be used by industry as a guide to predict yeast growth in wine
fermentations. To the best of our knowledge, this is the rst time that
this type of predictive model is used to compare the response of a
hybrid strain with respect to representatives of its parental species,
obtaining at the same time valuable information about the possible
effect of climatic change on yeast growth.
The effect of the environmental variables on was complex to
model due to the short lag period and the narrow interval in which
their maximum and minimum values were reported (7.49 h for yeast
T73, 9.21 h for hybrid W27, and 16.4 h for yeast 1802). Difculties in
estimating the bacterial lag time were also recognized by numerous
authors and some of them have reviewed the diverse modelling
possibilities (McMeekin et al., 1993; Swinnen et al., 2004). In the case
of B. intermedius i100, a yeast also isolated from grape must, was
reported a very short lag phase (7.80 h) in a medium with low pH
(3.50) (Medawar et al., 2003). These authors also found a high standard deviation during the estimation of the parameter (3.40 h).
Similar results were found in this work for Saccharomyces strains.
Models for max always had a non-signicant lack of t. The
temperature was the variable with the higher effects for the three
yeasts. This fact was also reported by Arroyo-Lpez et al. (2006) for
P. anomala, by Sorensen and Jakobsen (1997) for D. hansenii, and by
D'Amato et al. (2006) for S. cerevisiae. Serra et al. (2005) studied the
inuence of temperature and pH on the max of S. bayanus var. uvarum
P3 and S. cerevisiae VL3c, both yeasts related with wine fermentations.
They showed that the temperature was also the factor with the main
inuence on the response whereas the pH was less important. The
optimal temperature for S. bayanus var. uvarum growth was 30 C
when the pH was kept constant at 4.0, and the optimal temperature
for S. cerevisiae VL3c at pH 5.0 was 35 C. In this work, the highest max
value for yeast T73 was reported at 34.1 C and pH 4.76 (upper limits
established by the experimental region), which suggest that the
optimum value of temperature for this strain could be even higher.
Torija et al. (2003) studied the effects of temperature (1535 C)
on the growth of different strains of S. cerevisiae in white must.
Overall, the fermentation was faster at high temperatures (3035 C).
Charoenchai et al. (1998) also mentioned that growth rates for S.
cerevisiae, P. anomala, K. apiculata, and T. delbrueckii increased with
temperature, although their study only reached 25 C. Charoenchai
et al. (1998) reported that variation of medium pH between 3.0 and
4.0 did not signicantly affect the growth rate and cell biomass of wine
yeasts. However, Fleet and Heard (1993) observed that growth rate
and must fermentation by S. cerevisiae were decreased as the pH was
decreased from 3.5 to 3.0. In this work, an increase in the pH from 2.24
to 4.76 increased max for yeasts T73, W27 and 1802. On this respect, if
acidity decreases in grapes due to the climatic change, growth rate
may increase for Saccharomyces wine strains. In the present study the
pH of the synthetic must was adjusted with tartaric acid (pKa = 2.98).
Therefore, the undissociated form of this organic acid could also
have an additional effect on yeast growth in the lower ranges of
pH (for example, at pH 2.24 a great proportion of the acid is still
undissociated).
Another factor that should be considered with detail is the sugar
concentration because yeast can delay its growth at high concentrations of glucose and fructose, which was clear for strains T73 and the
hybrid W27 (Table 2). Sugar concentrations from 200 g/L to 300 g/L
decreased S. cerevisiae growth rate, as reported by Charoenchai et al.
(1998) and D'Amato et al. (2006). These authors found the lowest
growth rate at the higher glucose concentrations. Belloch et al. (2008)
studied the response of Saccharomyces hybrids in laboratory media

under low pH (2.8) and 250 g/L of glucose. The majority of yeasts
tested were able to grow at 30 C, but with the methodology used,
they could not detect small changes in their responses. Therefore, the
grape sugar concentration increase produced by the climatic change
could negatively affect the kinetics of the wine fermentations.
The response of the yeasts was different depending on the variable
under study. In the case of temperature, the response of the hybrid
W27 was similar to T73 showing a positive and linear effect when
temperature moved from 13.9 to 34.1 C, while yeast 1802 showed a
negative and quadratic effect with an optimum of temperature around
24.0 C. D'Amato et al. (2006) also reported a positive and linear effect
for a S. cerevisiae wine strain in the interval of temperature 1535 C.
These authors did not nd an optimum of temperature in the range
studied. However, when the studied factor was pH, the hybrid W27 had
a similar response to yeast 1802 and different to T73. The response of
the hybrid W27 as a function of the sugar concentration had a similar
behaviour to T73 and different to 1802. Serra et al. (2005) reported that
hybrid strains between S. bayanus var. uvarum and S. cerevisiae globally
had a response as a function of the temperature similar to the parental
S. bayanus var. uvarum, although not completely identical. In the
present work, the hybrid W27 response to temperature was also
similar to that of the parental S. cerevisiae representative (strain T73),
but showing a higher max (Table 2). The yeasts T73 and 1802 were
chosen as representative strains of the parental species involved in the
origin of hybrids, such as W27, because the real parental strains of
natural hybrids are unknown. This fact can explain the small
differences in the response of the hybrid with respect to the parental
representative. Gonzlez et al. (2007) showed that the fermentation
patterns of hybrid W27 and yeast T73 in Tempranillo musts at different
temperatures were also very similar, especially in the range of temperature between 18 and 32 C. Gonzlez (2005) performed a detailed
molecular characterization of the hybrid W27 using ow cytometry,
comparative genome hybridization with DNA macroarrays and RFLP
analysis of 37 genes located in the different chromosomal arms. Her
results showed that hybrid W27 is an aneuploid strain, with three
types of chromosomes, a complete set of the 16 chromosomes of
S. cerevisiae, most chromosomes of the S. kudriavzevii parental, and
several chimerical chromosomes resulting from recombination between homeologous chromosomes from S. cerevisiae and S. kudriavzevii. This complex genome structure could explain the response of the
hybrid with respect to the representative strains of the parental species
depending on the environmental variables under study.
5. Conclusions
Results obtained in this work show that changes in values of
fermentation temperature, pH and sugar concentration of the must
originated by the climatic change may affect to wine yeast growth. RS
equations built for T73 and the hybrid W27 can be used by the industry
to predict the growth of these yeasts under different combinations of
the environmental variables. S. kudriavzevii, which has never been
isolated from wine environments, showed optimal growth under low
pH, high sugar concentration and temperatures around 24 C, not
unusual during winemaking. This observation raises the question
about why this species in not found in wine environments, whilst its
hybrids with S. cerevisiae are found in Central European wine fermentations. Further studies on the interaction between these species
during fermentation are necessary to answer such a question.
Acknowledgements
This work was supported by the Generalitat Valenciana (project
GV2008-037) and the Spanish Government (projects AGL200612703-CO2-01 and 02/ALI). F.N. Arroyo-Lpez and S. Orli also want
to thank the Ministry of Education and Science of Spain (MEC) for
their respective postdoctoral research contracts.

F.N. Arroyo-Lpez et al. / International Journal of Food Microbiology 131 (2009) 120127

References
Arroyo-Lpez, F.N., Durn Quintana, M.C., Garrido Fernndez, A., 2006. Use of the
generalized z-value concept to study the effects of temperature, NaCl concentration
and pH on Pichia anomala, a yeast related to table olive fermentation. International
Journal of Food Microbiology 106, 4551.
Baranyi, J., Pin, C., Ross, T., 1999. Validating and comparing predictive models. International Journal of Food Microbiology 48, 159166.
Belloch, C., Orli, S., Barrio, E., Querol, A., 2008. Fermentative stress adaptation of
hybrids within the Saccharomyces sensu stricto complex. International Journal of
Food Microbiology 122, 188195.
Bely, L., Sablayrolles, J., Barre, P., 1990. Description of alcoholic fermentations kinetics:
its variability and signicance. American Journal of Enology and Viticulture 49,
319324.
Bradbury, J., Richards, K., Niederer, H., Lee, S., Rod Dunbar, P., Gardner, R., 2006. A
homozygous diploid subset of commercial wine yeast strains. Antonie van
Leeuwenhoek 89, 2737.
Charoenchai, C., Fleet, G., Henschke, P.A., 1998. Effects of temperature, pH and sugar
concentration on the growth rates and cell biomass of wine yeasts. American
Journal of Enology and Viticulture 49, 283288.
Coleman, M.C., Fish, R., Block, D.E., 2007. Temperature-dependent kinetic model for
nitrogen-limited wine fermentations. Applied and Environmental Microbiology 73,
58755884.
Colombi, S., Malherbe, S., Sablayrolles, J.M., 2005. Modeling alcoholic fermentation in
enological conditions: feasibility and interest. American Journal of Enology and
Viticulture 56, 238245.
D'Amato, D., Corbo, M.R., Del Nobile, M.A., Sinigaglia, M., 2006. Effects of temperature,
ammonium and glucose concentrations on yeast growth in a model wine system.
International Journal of Food Science & Technology 41, 11521157.
Fleet, G.H., Heard, G.M., 1993. Yeasts-growth during fermentation. In: Fleet, G.H. (Ed.),
Wine Microbiology and Biotechnology. Harwood Academic Publishers, Chur,
Switzerland, pp. 4243.
Gonzlez, S.S., 2005. El papel de nuevas especies de Saccharomyces y sus hbridos en
procesos de vinicacin. Doctoral Thesis. Department of Chemistry. Universidad
Politcnica de Valencia. Spain.
Gonzlez, S.S., Barrio, E., Gafner, J., Querol, A., 2006. Natural hybrids from Saccharomyces
cerevisiae, Saccharomyces bayanus and Saccharomyces kudriavzevii in wine fermentations. FEMS Yeast Research 6, 12211234.
Gonzlez, S.S., Gallo, L., Climent, M.D., Barrio, E., Querol, A., 2007. Enological characterization of natural hybrids from Saccharomyces cerevisiae and S. kudriavzevii.
International Journal of Food Microbiology 116, 1118.
Jones, G.V., White, M.A., Cooper, O.R., Storchmann, K., 2005. Climate change and global
wine quality. Climatic Change 73, 319343.

127

Malherbe, S., Fromion, V., Hilgert, N., Sablayrolles, J.M., 2004. Modeling the effects of
assimilable nitrogen and temperature on fermentation kinetics in enological
conditions. Biotechnology and Bioengineering 86, 261272.
McMeekin, T.A., Olley, J.N., Ross, T., Ratkowsky, D.A., 1993. Predictive Microbiology:
Theory and Application. John Wiley & Sons, Inc, New York.
Medawar, W., Strehaiano, P., Dlia, M.L., 2003. Yeast growth: lag phase modelling in
alcoholic media. Food Microbiology 20, 527532.
Myers, R.H., Montgomery, D.C., 2002. Response Surface Methodology, 2nd ed. John
Wiley & Sons, Inc, New York.
Naumov, G.I., James, S.A., Naumova, E.S., Louis, E.J., Roberts, I.N., 2000. Three new
species in the Saccharomyces sensu stricto complex: Saccharomyces cariocanus,
Saccharomyces kudriavzevii and Saccharomyces mikatae. International Journal of
Systematic and Evolutionary Microbiology 50, 19311942.
Pretorius, I.S., 2000. Tailoring wine yeast for the new millennium: novel approaches to
the ancient art of winemaking. Yeast 16, 675729.
Querol, A., Fleet, G., 2006. Yeasts in Food and Beverages. Springer-Verlag, Berlin,
Germany.
Querol, A., Huerta, T., Barrio, E., Ramn, D., 1992. Dry yeast-strain for use in fermentation
of Alicante winesselection and DNA patterns. Journal of Food Science 57, 183.
Rossignol, T., Dulau, L., Julien, A., Blondin, B., 2003. Genome-wide monitoring of wine
yeast gene expression during alcoholic fermentation. Yeast 20, 13691385.
Schtz, M., Gafner, J., 1994. Dynamics of the yeast strain population during spontaneous
alcoholic fermentation determined by CHEF gel electrophoresis. Journal of Applied
Bacteriology 19, 253257.
Serra, A., Strehaiano, P., Taillandier, P., 2005. Inuence of temperature and pH on Saccharomyces bayanus var. uvarum growth; impact of a wine yeast interspecic
hybridization on these parameters. International Journal of Food Microbiology 104,
257265.
Sorensen, B.J., Jakobsen, M., 1997. The combined effects of temperature, pH and NaCl on
growth of Debaryomyces hansenii analyzed by ow cytometry and predictive
microbiology. International Journal of Food Microbiology 34, 209220.
Swinnen, I.A.M., Bernaerts, Kdens, E.J.J., Geeraerd, A.H., Van Impe, J.F., 2004. Predictive
modelling of the microbial lag phase: a review. International Journal of Food
Microbiology 94, 137159.
Torija, M.J., Rozs, N., Poblet, M., Guillamn, J.M., Mas, A., 2003. Effects of fermentation
temperature on the strain population of Saccharomyces cerevisiae. International
Journal of Food Microbiology 80, 4753.
Zwietering, M.H., Jongenburger, I., Rombouts, F.M., Van't Riet, K., 1990. Modeling of the
bacterial growth curve. Applied and Environmental Microbiology 56, 18751881.